English

DeepShift: Towards Multiplication-Less Neural Networks

Machine Learning 2021-07-09 v5 Neural and Evolutionary Computing

Abstract

The high computation, memory, and power budgets of inferring convolutional neural networks (CNNs) are major bottlenecks of model deployment to edge computing platforms, e.g., mobile devices and IoT. Moreover, training CNNs is time and energy-intensive even on high-grade servers. Convolution layers and fully connected layers, because of their intense use of multiplications, are the dominant contributor to this computation budget. We propose to alleviate this problem by introducing two new operations: convolutional shifts and fully-connected shifts which replace multiplications with bitwise shift and sign flipping during both training and inference. During inference, both approaches require only 5 bits (or less) to represent the weights. This family of neural network architectures (that use convolutional shifts and fully connected shifts) is referred to as DeepShift models. We propose two methods to train DeepShift models: DeepShift-Q which trains regular weights constrained to powers of 2, and DeepShift-PS that trains the values of the shifts and sign flips directly. Very close accuracy, and in some cases higher accuracy, to baselines are achieved. Converting pre-trained 32-bit floating-point baseline models of ResNet18, ResNet50, VGG16, and GoogleNet to DeepShift and training them for 15 to 30 epochs, resulted in Top-1/Top-5 accuracies higher than that of the original model. Last but not least, we implemented the convolutional shifts and fully connected shift GPU kernels and showed a reduction in latency time of 25% when inferring ResNet18 compared to unoptimized multiplication-based GPU kernels. The code can be found at https://github.com/mostafaelhoushi/DeepShift.

Keywords

Cite

@article{arxiv.1905.13298,
  title  = {DeepShift: Towards Multiplication-Less Neural Networks},
  author = {Mostafa Elhoushi and Zihao Chen and Farhan Shafiq and Ye Henry Tian and Joey Yiwei Li},
  journal= {arXiv preprint arXiv:1905.13298},
  year   = {2021}
}

Comments

-Added results for 8-bit and 16-bit fixed point activations, as well as 5-bit, 4-bit, 3-bit, and 2-bit weights. - Added link to GitHub code - Updated and fixed the training algorithm - Introduced 2 approaches for backward and forward pases - Showed better results for training from scratch on CIFAR10 and Imagenet - Added implementation on NVIDIA's GPU -Accepted in CVPR Mobile AI 2021 Workshop

R2 v1 2026-06-23T09:34:04.418Z